Conventionally, the following methods, for example, are known for connecting a pipe to another member.
A pipe is inserted into an insertion hole formed in another member, and a split die circumferentially split into multiple segments is disposed inside a hollow portion of the pipe. Next, by moving each segment of the split die radially outward of the pipe, the insertion portion of the pipe inserted in the insertion hole and both portions of the pipe axially adjacent to the insertion portion are subjected to expansion work (tube diameter expansion work) to thereby join the member to the pipe (for example, see Patent Documents 1 and 2). This method is also called a ridge lock method.
Although it is not a joining method for joining members, as an expansion method for expanding a tubular member using a split die, a method described in the Japanese Unexamined Laid-open Patent Publication No. 2002-224743, for example, is known (see Patent Document 3).    Patent Document 1: Japanese Unexamined Laid-open Patent Publication No. H4-8818 (page 2, FIG. 8)    Patent Document 2: Japanese Unexamined Laid-open Patent Publication No. H11-36859    Patent Document 3: Japanese Unexamined Laid-open Patent Publication No. 2002-224743